* [Warren v0.18.0](https://github.com/jayminwest/warren) – Self-hostable control plane that spawns ephemeral, sandboxed coding agents on GitHub repos, streams events, and returns a branch for PRs.

* [Strands Agents python/v1.53.0](https://github.com/strands-agents/harness-sdk) – Model-driven SDK for building and running AI agents, with customizable agent loops, multi-agent workflows, and model-agnostic providers.

* [OpenSwarm v0.21.0](https://github.com/unohee/OpenSwarm) – Autonomous orchestrator that runs Claude Code agents to process Linear issues, produce code changes, report to Discord, and store memory in LanceDB embeddings.

* [REDCELL v0.2.0](https://github.com/martian56/redcell) – REDCELL orchestrates LLM agents to execute real penetration testing in a Kali container and generate structured reports with live operator control.

* [Brigade brigade-v1.30.0](https://github.com/spinabot/brigade) – Self-hosted ecosystem of AI agents on an org chart with shared long-term memory, deployable via terminal and messaging clients while keeping keys and data local.

* [AI Maestro v0.36.43](https://github.com/23blocks-OS/ai-maestro) – Agent orchestrator providing persistent memory, agent-to-agent messaging, and a dashboard to manage agents across multiple machines.

* [LobsterAI 2026.8.18](https://github.com/netease-youdao/LobsterAI) – 24/7 personal assistant agent that executes tasks, automates workflows, runs code locally or in a sandbox, and supports IM control.

* [ReevesAgents v1.7.6](https://github.com/mertkayacs/reevesagents) – Local tmux-first workspace that lets one AI coding agent spawn, steer, and supervise other agent CLIs via MCP with shared TUI, Web UI, and CLI control.